In inclusive P-wave charmonium decay, to cancel infrared divergence in the colour singlet contribution requires the inclusion of the colour octet which becomes degenerate with the former in the infrared limit. In the corresponding exclusive decay, such an infrared divergence does not exist. On this ground, it becomes doubtful whether the colour octet is needed in exclusive reactions. A contradiction in the underlying picture, however, would result once all strong decay channels are summed. We provide an answer to this question in support of the colour octet with theoretical arguments as well as an explicit calculation of Chi_J ---> p \bar p.